Electronic record-keeping produces data sets with thousands or millions of records. The sheer quantity of information available for analysis may prevent meaningful conclusions from being drawn, or may prevent analysis from even being attempted. Data items in data sets may be associated with probabilities or risks that can be aggregated across various item dimensions, and an analysis of the probabilities associated with a particular data item may be incomplete or misleading without further consideration and analysis of aggregate probabilities.